Two Paradigm Shifts: Trust Infrastructure and AI-Agentics

For the past seven years, I have been advocating for faster adoption of organisation wallets to enable verifiable credentials to start streaming through the European Trust Infrastructure (TI). Much of this has been in support of the public–private not-for-profit Findynet co-operative in Finland, and as a co-founder or active contributor in initiatives like Mobey Forum, Real Time Economy, MyData, Trust over IP, and the Open Wallet Foundation.

From the start, the why has been clear: dramatic leaps in
• productivity (3–6% GDP growth potential),
• security and risk management,
• privacy and citizen empowerment,
• service quality in the public sector and enterprises
• cybercrime/hybrid warfare threat resilience, and
• a more efficient European Single Market.

But in 2023–2024, something even bigger emerged.

The arrival of ChatGPT-4 and its peers signaled a second paradigm shift. As AI learned to speak, we realized we were entering not just a continuation of the TI journey—but an entirely new chapter. One where AI-agents become deeply embedded in everyday workflows, representing people and organizations, communicating with each other, and eventually coordinating with shared goals.

Crucially, these AI-systems can only function safely and effectively if they are anchored in trust. That’s where the verifiable credentials of the Trust Infrastructure—issued and held in ID-wallets—become indispensable.

In this sense, the two revolutions—TI and AI-agentics—are not parallel but interdependent.

I’ve been engaging in daily discussions on LinkedIn and weekly meetings with stakeholders across the world, while also continuing to read broadly. One highlight: I just finished Jonnie Penn Summerfield’s excellent These Strange New Minds. Though the manuscript was ready in late 2023, his November 2024 afterword captures the shockwave: “the rate at which new ideas and results have emerged has been simply mind-bending.”

The book elegantly traces the emergence of thinking, deep learning’s superhuman feats, and the philosophical and practical questions around what LLMs know, say, and do. Summerfield anticipates the arrival of personalized AI-agents before that term had gained traction:

“This diversity [of AI-systems] is coming as a natural consequence of our desire for LLMs that are personalized and trained to know each of us intimately. AI personalization will augur a new era in which we are accompanied by a melting pot of different LLM-grounded agents, each acting as a personalized assistant – a digital representative for a person or a group.”

What was not yet fully present in his framework—unsurprisingly for late 2023—was the central role of the Trust Infrastructure and identity-building wallets as essential risk mitigators.

But that’s changing—fast.

The future won’t be either TI or AI. It will be TI–powered AI.
AI-agents won’t gain real traction until they can interact legally and securely—which means they must operate through the same verifiable trust fabric that people and organizations already do.